Evaluation of the SONAR Meter in Wet Gas Flow for an Offshore Field Development Anela Floyd, BP Siddesh Sridhar and Gabriel Dranea, Expro Meters 1 INTRODUCTION The ABC project is a hih pressure as condensate development currently in the implementation stae. Platform top-side surveillance is considered critical for better reservoir manaement, production optimization and flowline interity manaement. After investiatin different options, the SONAR clamp-on meter was identified as the solution to provide individual, real time production surveillance for ten flow lines (12-inch pipe) upstream of the production manifolds and the hih-pressure (HP) separator. In order to determine the wet as flow uncertainty over the anticipated as and liquid flow rane, testin was performed at the NE, Wet Gas Facility in Scotland. The results and conclusions from the testin are detailed in this paper. The main requirement of the test proram was to evaluate the SONAR meter for provision of real time as rates and inferred oil and water rates across a rane of flow conditions anticipated across the life of hih pressure as condensate development. This testin was limited to lab capabilities of; Measure max pressure of 63 Bar at 2 flow rates of 780 ACMH and 1040 ACMH 0 10% WR Volumeteric Gas rate accuracy 5-10% Volumetric iquid Rate Accuracy 10-15%. Mass Bulk flow accuracy dependent on density measurement 2 SONAR Technoloy SONAR technoloy involves observin the naturally-occurrin coherent vortical structures within the flow (enerated due to turbulence) by monitorin interactions of externally-enerated acoustic pulses (pulse arrays) with those coherent structures. The subsequent processin alorithms involve analysis of the spatial wavelenth (distance) and temporal frequency (time) of the sensor sinals over a rane of values. Multiple spatial and temporal wavelenths are plotted to enerate a k-omea plot which is essentially a hih enery reion called the vortical ride. The slope of this ride determines the flow velocity. The volumetric flow rate at standard conditions can then be calculated usin the pipe cross-sectional area, pressure and temperature. 1
Flow Velocity: 12.6 ft/s Fi.1 - SONAR Dianostics: k-ω ride 3 SONAR OVERREADING CORREATION - BACKGROUND In wet as flow measurement, the flow meter enerally over reads the as flow rate due to wetness addition. The over-readin is defined as OR = (1) ref where is the as volumetric flow rate measured by the flow meter and the actual as volumetric flow rate iven by the reference system in line condition. Based on empirical data collected in flow loop testin conducted on 4-in Sch 40 and Sch 80 pipes, the over-readin of the 4-in ActiveSONAR TM flow meter can be characterized by the the followin correlation: VF OR 1+ 1+ Fr where β = 2. 5249, ϕ = 3. 9043 and VF + ϕ 1 + Fr = m m F = 1, The F r number is defined as 2 ref β (2) 0.5, 0.5 < F r 1.5 < 1.5 r r m (3) F r V s F ρ = (4) ρ ρ where V s is the superficial velocity of the as flow, the ravitational acceleration, the characteristic lenth of the pipe (ID of the pipe here), ρ the as density in line conditions and ρ the liquid density in line conditions. 2
The iquid Volume Fraction ( VF) is defined as VF + = (5) where is the liquid volumetric flow rate in line conditions. For low liquid loadin, << to the as volumetric rate, Eq. (5) is simplified as the ratio of the liquid volumetric rate VF (6) The over-readin correction shown in Eq. (2) enables the 4-in ActiveSONAR TM meter to report as rates to within ±3% for 0<VF<0.106 and 0.5< F r <5.78. There is another important dimensionless parameter for liquid fraction broadly used in wetness flow meterin is ockhart-martinelli number, which is defined as X For low liquid loadin, we have ρ M = (7) ρ X M VF ρ ρ (8) 4 TEST SETUP Testin was performed in the NEl hih pressure wet as facility in 2 phases, Oct- 2014 and Mar-2015. The flow meter was clamped onto a 12 sch 160 pipe spool with class 600 raised face flanes. The meter was mounted approximately 14.5 D (4.3m) downstream from the 12-in straiht pipe upstream flane face. This setup was used in order to replicate the expected installation setup at the platform conditions. The test section pressure tappin was located 1.5 m from the downstream flane and the temperature probe was located 643 mm from the downstream flane. The flow computer was installed on a vertical ratin adjacent to the meter location. An identical test setup was used for both phases of testin. Phase 1 testin used Nitroen as via 6 ultrasonic meters and kerosene via ½, 1 and 3 turbine meters. Testin followed a defined test matrix and maintained a ~63 bar() pressure, 15 C and as volumetric flow rates of 780 1040 m 3 /hr. Durin the 6 month time from Phase 1 - Phase 2 testin, the NE lab had been updated to a multiphase flow lab and incorporated the use of a 3 and 1 1/2 Coriolis meter as the liquid kerosene reference meters. The 6 USM remained the as reference meter and maintain a ~63 bar() pressure, 15 C and as volumetric flow rates of ~500 1600 m 3 /hr. The flow was wet as, 100% nitroen mixed with kerosene oil. After the eometric parameters of the pipe were obtained, the flow meter was confiured. The drivin frequency of the transducers and demodulation delay were obtained by usin the built-in oprimizer. The confiuration of the flow meter was fixed 3
durin the whole test period and identical confiurations were used for both rounds of testin. The line pressure and temperature for both phases of testin raned from 60-62 bar and 15-20 0 C respectively. The flow rate, pressure and temperature were stabilized before a 30-second pretest point was loed. The CGR (condensate-as ratio) for each test point was then provided to Expro for input into the flow computer. The actual test point was then loed for a period 240 seconds (4 minutes). Meter performance criteria was defined as 5% to 10% for the as flow rate measurement and 10% to 15% for liquid flow rate measurement. 5 TEST RESUTS 5.1 1st Testin Phase Testin was conducted for 65 data points in total with 5 dry as points and 60 wet as points. For the dry as test points, 2 points were tested before the wet as test points and 3 points were tested after the wet as test points. The testin was conducted at as flow rates of 620 and 800 m3/h (at actual conditions). The Froude Numbers (Fr) rane for the test points raned from 0.69 to 0.89 and the iquid Volumetric Fraction (VF) raned from 0 10.7%. After NE reference data was obtained, a comparative analysis was done between Expro and NE as and liquid rates for all the test points. For dry as test points, the difference in as at actual conditions was less than +1% which is within the SONAR meter specifications as shown in Fiure 2. Fi. 2 - Dry Gas Error (SONAR vs. NE) 1 st Testin Phase 4
For the wet as test points, the averae difference between Expro as (at Actual conditions) and NE Reference as (Actual) was approximately -15% (Fiure 3). Upon further investiation this error was attributed primarily due to the overreadin correlation. A secondary source of error was a mismatch in CGR (condensate as ratio) for some of the test points and the actual CGR at flowin conditions. A modified over-readin correlation was developed and this has been detailed in section 5.2 Fi. 3 - Wet Gas Error (SONAR vs. NE) 1 st Testin Phase 5.2 Modified Overreadin Correlation For the first phase of testin, the over-readin correlation for 4-in wet as flow shown in Eq. (2) and Eq. (3) (incorporated in the flow computer) was applied to the 12-in wet as test points. Fi. 4-1 compares the actual over-readin for the NE test points (OR NE) and the reported over-readin usin the existin 4 correlation (OR SONAR). The error in as rate has been plotted on the Y-axis. It is evident that the actual over-readin is considerably lower than the 4 overreadin correlation and follows a linear trend. This indicates that the over-readin correction is pipe size dependent. Since the oriinal correlation was developed for a 4 pipe and the NE testin was conducted on a 12 pipe, the meter overreadin seems to decrease as the pipe size is increased. 5
Fi. 4-1SONAR over readin versus VF (1 st Testin Phase) Upon further analysis of the data, it was determined that the over-readin for the 12-in wet as flow can be characterized by modifyin the coefficients of Eq. (2). After a curve-fittin exercise, the followin correlation was obtained for the 12-in wet as flow: VF VF OR = 1+ + 1 ϕ m m 1 + Fr + Fr where β = 0. 037467, ϕ = 6. 168157 and { F r 0.5, 0.5 < F < 1. 1 β (12) m (13) = r The above correlation has been developed only up to Fr = 1.1 and VF < 8%. Additional reference data was needed to validate the correlation at hiher Froude numbers. The modified over-readin correlation was then applied to all the test points. Fi. 2 shows the curvin fittin of the test points. 2 6
Fi. 2 - Over-readin curve-fittin for 12-in meter in wet as flow Error! Reference source not found. shows the as error (SONAR vs. NE) after applyin the modified over-readin correlation to the test data. As evident from the raph, the difference between the SONAR as rates and NE as rates was within ±2% for most test points. Fi. 6 - SONAR OR and as Error versus VF C-Fit vs. Oriinal OR 7
5.3 2 nd Testin Phase The 2 nd round of testin at NE was conducted in March 2015. The primary purpose of the test was to validate the over-readin correlation developed for the 12 meter based on the data analysis after the 1 st round of testin. The new overreadin correlation was implemented in the flow computer prior to the commencement of the test. As stated earlier, the meter and flow computer setup was identical to the previous setup in order to maintain repeatability of operatin conditions. The test matrix was however condensed to 43 test points, 6 dry as points and 37 wet as test points. The as (actual) rane for the test points was 500-1000m3/hr. The Froude Numbers (Fr) rane for the test points raned from 0.69 to 1.1 and the iquid Volumetric Fraction (VF) raned from 0 10.2%. Fiure 7 shows the comparison between the actual over-readin for the test points and the reported over-readin usin the modified correlation usin curvefittin. It is evident from the chart that the modified over-readin correlation follows the same trend as the actual over-readin with increasin VF. Fi. 7 - SONAR over readin versus VF (2 nd Testin Phase) Fiure 8 shows the comparison between SONAR as actual and NE reference as actual versus VF. As can be seen in the chart, 93% of the SONAR reported as rates were within + 5% of the reference rates. The remainin points were within 10% of the reference rates. The liquid rates are directly inferred from the as rates usin the CGR and hence are within the same error bands (Fiure 9). 8
Fi. 8 - as error versus VF (2 nd Testin Phase) Fi. 9 - liq error versus VF (2 nd Testin Phase) 9
6 CONCUSIONS The meter qualification testin performed in the 1 st phase (Oct-2014) led to the development of a modified over-readin correlation for the 12-inSONAR meter in wet-as flow. This correlation was also validated in the NE test loop by additional testin in the 2 nd Phase (March-2015). It is evident from the testin at NE (and previous flow loop testin) that the SONAR meter over-readin characteristic for wet as flows is pipe size dependent. The over readin also seems to decrease with increasin pipe size. It is recommended to perform testin at intermediate pipe sizes (6 and 8 ) and at hiher Froude numbers to characterize the meter over-readin and subsequently implement it for future field applications. 7 NOTATIONS β Beta (Calibration Coefficient) ϕ Phi (Calibration Coefficient) VF iquid Volumetric Fraction OR Overreadin C-Fit Curve-Fit CGR Condensate-Gas Ratio Fr Froude Number 8 REFERENCES [1] Shoham, O., Mechanistic Modelin of Gas-liquid Two-Phase Flow in Pipes, Society of Petroleum Enineers, 2006, ISBN 978-1-55563-107-9 [2] Murdock, J. W., Two Phase Flow Measurement with Orifices, Journal of Basic Enineerin, Vol.84, pp.419-433, 1962 [3] Konopczynski, M.R., are-scale Application of Wet-Gas Meterin at the Oman Upstream NG Project, SPE 63119, presented at the 2000 Annual Technical Conference in Dallas, TX, USA, October, 2000 [4] Gyslin, D. and Morlino, N., Pulsed-Array, Sonar-based Flow Measurement Technoloy for Clamp-on Wet Gas Meterin, Presented at the 2009 Americas Flow Measurement Workshop, Houston, TX, USA February, 2009 [5] Danesh, A., PVT and Phase Behaviour of Petroleum Reservoir Fluids, Developments in Petroleum Science, Vol. 47, Elsevier Science B.V., ISBN: 0-444-82196-1, 1998 [6] Gyslin D., u M. and Wen T., Clamp-on Two Phase Measurement of Gas Condensate Wells Usin Interated Equation of State Compositional Models, 28 th International North Sea Flow Measurement Workshop, October, 2010 10